RoboComp Logo

A simple robotics framework.

GSoC 2021 RoboComp An object detection component for LearnBlock

18 June, 2021

About me

Hi, my name is Alejandro Fernández Camello and I am studying a degree in Software Engineering at the University of Extremadura. This is my first participation in GSoC and I will try to do my best.

  • My interests: Deep Learning, Reinforcement learning, Cryptography and Data Structures.

About the project

The goal of the project is to extend the computer vision functionality of LearnBlock adding a component of object detection. The model that is going to be used needs to work in a low computational environment and be capable of detecting a reasonable number of objects, these requirements will guide the development of the project. The dataset will be COCO that can detect 90 type of objects and respecting to the model we have three differents options, Tiny-Yolo, Mask R-CNN and SSD MobileNet, of these we will select the model with better features for the project. After that, the model will be converted to the format of Tensorflow Lite in case is necessary and finally the model will be implemented in LearnBlock.

Alejandro Fernández Camello